Cone Testing

Cone tests require a square specimen ~200x200mm. The bead on plate welds created here are far too small for this test. Bead on plate welds have been recreated in a butt weld configuration with specimens measuring 200x330mm. These can be successfully cone tested but throw up many more issues with the setting up of the welds and general success of the welds. The butt configuration re-introduces the plunge depth issues highlighted in the Parkson milling machine welds. The recreated

101

welds used the cold welding parameters (500Rpm, 746mm/min and 1.49 mm/rev) and have been carried out using the 5651 tool (7.5 mm diameter), a simple concave tool (16mm diameter) and the large shoulder tool (19mm diameter). The setting up of the welding process is hampered by the manual plunge depth of this machine. Welds have been created and immediately tested for quality using the 180° bend testing with the root in tension. Although the welds passed the bending test it was feared that these welds would not survive the elevated temperature cone tests and would rupture when exposed to the gas pressures used during forming. One weld created by each tool was selected for a multiple pass welding trial where the FSW was carried out in two consecutive passes on the top surface and then again on the bottom surface. Both of the welding passes were carried out using the cold welding parameters (500rpm, 746mm/min) in order to minimise the heat input to the weld. This makes sure that there is sufficient bonding in the through thickness of the material and the removal of root defects so the welded specimen is able to withstand the forming process.

Table 7.6 shows the results of the elevated temperature cone test results for the welds created using FW22. Figure 7.12 shows pictures of the resulting cone test specimens.

It is immediately clear that the single pass welds do not fare well in this test.

Although more single pass welds were tested the two shown in Table 7.6 are actually the most successful of the single pass welds. The other omitted welds could not achieve gas pressure and so the test could not be completed. This further highlights the significant effects of root flaws in friction stir welds. The single pass welds have proven successful in bend testing but this is a uniaxial test pulling the weld root apart. Cone testing uses gas pressure to apply biaxial stress in the material. Any flaw in the weld root will be immediately exploited by the gas causing the weld seam to rupture. The double pass welds have been much more successful in maintaining pressure but there is still a major difference in success between the two materials used.

102

Table 7.6, Cone Test Results for Welds Created using FW22.

Figure 7.12, Cone Tested Specimens of Welds Created using FW22.

a) AA2004 – Double Pass (19mm tool), b) AA5083 – Double Pass (19mm tool), c) AA5083 – Double Pass (7.5mm tool), d) AA5083 – Single Pass (7.5mm tool),

e) AA5083 – Single Pass (16mm tool).

103

The most significant aspect of this test is that only the AA2004 weld survived the gas pressure applied during the test. This weld shown Figure 7.12.a) formed to an equivalent superplastic elongation of ~260%. This specimen has a completely different failure mode to the other AA5083 specimens. This failure mode is grain coarsening. During SPF the material experiences heat and deformation, this gives the material energy for microstructural change and so the material goes through recrystallization, recovery and grain growth mechanisms. The friction stir weld in these specimens is an introduced heterogeneity in the parent material. The differing structure of the weld causes the recrystallization, recovery and grain growth mechanisms to act differently on this material compared to the surrounding parent material. It can be seen from the photo in Figure 7.12 a) that the weld material has begun coarsening well before the parent material to the extent that the grain structure of the FSW has grown to such proportions that individual grain can be seen with the naked eye; this effect is sometimes referred to as the ―orange peel‖ effect.

The AA5083 welds created for this test were unable to survive the SPF process in both a single and double pass operation. The smallest tool (7.5mm) was tested in both single and double pass operations, but both failed in the same way; weld rupturing when subjected to gas pressure. AA5083 is commercially available structural aluminium alloy which is specially processed to provide a microstructure capable of superplastic elongations. This level of superplasticity is far less than AA2004 which is a specifically designed SPF alloy. In the case of the AA2004 it seems that the weld material begins to deform before the parent material causing the weld material to stretch with the parent material constraining this deformation at the extremities of the weld; this can be seen in Figure 7.12.a) where the deformation is localized within the weld region. For the AA5083 however it is evident that the weld material has become undeformable under these conditions in a similar way to conventional fusion welds [80]. The parent material begins to deform first but the weld material resists the deformation and is pulled apart by the stretching parent material.

FW22 has successfully created FSWs in aluminium which can be subjected to subsequent SPF processes. However, every pass completed by the FSW tool changes the microstructure so a double pass weld will have experienced more disruption and

104

more deformation and heating phenomena than a single pass weld. This machine was unable to create successfully formed welds in a single pass operation. This is again due to poor bonding in the weld root and an inherent flaw in the weld root. The FW22. The use of this machine has also enabled faster spindle and traverse speeds to be used. The welding pitch has remained close to the values used for FW22 but both the spindle speed and feed speed have been increased. This increase in spindle speed is based on experimental data from TWI which shows that 2XXX series aluminium alloys respond better to higher spindle speeds [30]. To prevent the higher spindle speeds from completely overheating and destroying the SPF microstructure the welding speed was also increased.

7.4.3.1 Tensile Test Results

The first most noticeable fact from the ESAB tensile results is that the ultimate tensile strength of the butts welds made on the ESAB machine are very similar to the strength of the bead on plate welds created on FW22 given the same size tools, this is shown in Figure 7.13. The 5651 (7.5mm) tool was not used on the ESAB machine but it is hypothesised that the results would be similar to those produced as bead on plate joints on FW22. It can also been seen that the welds produced resemble the strength of annealed material indicating further that the FSW process acts as a localized annealing process in the wake of the tool as it traverses the joint line.

The ESAB SuperStir machine can effectively reduce the chances of plunge depth related flaws by using position control. The programmable nature of the machine

105

means that once satisfactory set up; welds produced to the same quality can be assured for subsequent welds. Two tools were used on this machine, the simple concave tool Concave005, and an MXTriflute^TM tool provided and modified by TWI.

The MXTriflute^TM tool is 15mm in diameter, has a concave shoulder profile and 5mm diameter, threaded and fluted probe. The comparison between the tools sheds light on the plunge depth issues experienced before. Even with the dramatically improved accuracy of the plunge depth, the few welds created on the ESAB machine using the simple concave tool still exhibit flaws in the weld root and fail the bending tests. This can now be attributed to the simple and smooth profiles of the shoulder and probe for this tool. The lack of threads and re-entrant features on the probe limits the amount of deformation occurring in the weld root, poor interface breakup results and poor joint properties are inevitable. Swapping this smooth tool for the MX TriFlute tool instantly improves results. The welds pass the bending test and provide tensile strengths which are similar to bead on plate results. Tensile test results for the welds created used the MXTriflute^TM tool are shown in Figure 7.13.

Figure 7.13, Ultimate Tensile Strength Versus Heat Input for Welds Created on the ESAB SuperStir FSW Machine: a) AA5083, b) AA2004.

106

Triple pass welds have also been created using the ESAB machine. For these welds all three passes are carried out on the top surface, the passes are spaced 5mm apart which coincides with the diameter of the tool‘s probe. This has created welds with a tool footprint of 25mm, this larger footprint means more FSW material and will act as a friction stir processed material. Figure 7.14 shows a schematic diagram of the overlapped pass configuration.

Figure 7.14, Schematic Diagram of the Overlapped Passes of the Multiple Pass FSWs Created on the ESAB SuperStir machine.

7.4.3.2 Cone Testing

Cone testing has been carried out on all the welds completed using the ESAB SuperStir machine. All the specimens have been tested with a gas pressure of 0.138 MPa (20 psi) and a temperature of 460°C. Figure 7.15a-e) shows the results for the cone tests of butt welds created in AA5083 on the ESAB machine.

107

Figure 7.15, AA5083 Cone Test Results (ESAB SuperStir).

a) Single Pass: 1250 Rpm/1250 mm/min, b) Single Pass: 1250 Rpm/1000 mm/min, c) Single Pass: 1250 Rpm/750 mm/min, d) Single Pass: 750 Rpm/750 mm/min,

e) Triple Pass: 750 Rpm/750 mm/min.

108

Table 7.7, AA5083 Cone Test Results.

The modes of failure and forming time for the AA5083 welds are shown in Table 7.7. It is clearly visible that all of the single pass welds share the same failure mode.

The weld material is heterogeneous this means that there will be differing levels of structure coarsening within this region. This is highlighted by Figure 7.16.a) where the weld region has separated, parallel to the welding direction, clearly defining the nugget and TMAZ regions. The nugget consists of recrystallized material; the TMAZ consists of partially recrystallized material, this means that the weld region will coarsen at different rates. This has led to the failure of the specimen at the nugget/TMAZ interface. This is the same for all the single pass welds. The triple pass weld however, has a different failure mode. This weld has failed perpendicular to the weld direction which may be related to the semi-circular bands of material characteristic to the FSW process. Figure 7.16.b) shows a detailed photo of the AA5083 triple pass weld.

109

Figure 7.16, Detail of the Failure Mode for Cone Tested AA5083 Specimens:

a) Single Pass 1250rpm/750mm/min, b) Triple Pass 750rpm/750mm/min.

The triple pass FSW has undergone three overlapping friction stir welding processes.

This means that pass 1 has experienced more heat and deformation from the successive FSW passes, pass 2 has also experienced the extra heat and deformation from pass 3. This has produced a weld region which is 25mm wide with a weld nugget ~15mm wide. Each of the passes has affected the weld region and accounts for the accumulated deformation, heat and ultimately softening of the weld material.

The tensile results for the triple pass butt weld (750rpm/750mm/min) show that this weld material has undergone significant softening. The welding process has successfully annihilated the joint interface preventing the FSW microstructure from failing parallel to the welding direction. It is visible from Figure 7.16 that the failure is due to the coarsening of the weld region, this in turn may be related to the FSW‘s banded microstructure, perpendicular to the weld direction and situated in the weld centre between passes 2 and 3.

The friction stir welds in AA2004 react to the cone test in a very different way. The weld material deforms more readily than the parent material and has undergone superplastic deformation. In each cone test specimen the failure site is located on the advancing side of the weld nugget further highlighting the heterogeneous nature of friction stir welded microstructure. The triple pass weld did not complete the cone test to failure; so it is unsure what the final failure position and extent of the superplastic deformation would be. Figure 7.17 a-e) shows the AA2004 cone tested specimens of the butt welds created on the ESAB machine, with inset pictures of the

110

failure location. The first most noticeable point is the similarities between all of the AA2004 samples. It can be seen that the weld regions have all undergone extensive deformation, with the weld region stretching more than the parent material. The next most significant point is the location of the failure. Each specimen, with the exception of the triple pass weld, failed on the advancing side of the weld region parallel to the welding direction. This suggests coarsening of the weld microstructure and eventual separation of the interface between the weld nugget and TMAZ regions.

The nugget and TMAZ have coarsened at slightly different rates during the process due to heterogeneous microstructures created during FSW. It should also be noted that none of the cone tested welds failed alone the original joint interface.

111

Figure 7.17, AA2004 Butt Welded Cone Test Results (ESAB SuperStir) a) Single Pass: 1250 Rpm/1250 mm/min, b) Single Pass: 1250 Rpm/1000 mm/min, c) Single Pass: 1250 Rpm/750 mm/min, d) Single Pass: 750 Rpm/750 mm/min,

e) Triple Pass: 750 Rpm/750 mm/min.

Table 7.8 summarizes the AA2004 butt welded cone test results. The most successful weld has been carried out in a single pass operation using a welding pitch of 0.8mm/rev. This weld has produced an equivalent superplastic strain of 210%. There is no real trend between the weld pitch and the resulting SPF strain. However it can been seen that increasing the spindle speed from 750rpm to 1250rpm, whilst maintaining the weld pitch, improves the results and enable SPF strain of approaching 200%.

112

Table 7.8, AA2004 Butt Weld Cone Test Results.

The bead on plate experiments carried out on the ESAB SuperStir machine reveal the same failure locations as the butt welded specimens; parallel to the weld direction on the advancing side of the weld. Further supporting the notion that the failure is a result of differing rates of microstructural coarsening between the weld nugget and neighbouring TMAZ; causing the failure at the interface between the two regions.

Figure 7.18 a-d) shows the difference between the bead on plate single and triple pass welds in an as-welded and surface finished condition. The scalloping marks have been removed from for the surface finished specimen and the top surface smoothed and polished to a 9µm finish. Both the single and triple pass welds experienced a very small drop in performance for the surface finished specimens.

This drop in performance is not considered to be significant. Surface finishing can therefore be carried out between the FSW operation and the subsequent SPF operation to remove the characteristic scalloping marks left by the process and enabling a high class surface finish for the formed material.

113

Figure 7.18, AA2004 Bead on Plate Cone Test Results.

a) Single Pass: 1250 Rpm/1000 mm/min-As welded, b) Single Pass: 1250 Rpm/1000 mm/min-Surface Finished, c) Triple Pass: 1250 Rpm/1000 mm/min-As welded, d)

Triple Pass: 1250 Rpm/1000 mm/min-Surface Finished.

Table 7.9 shows a summary of the bead on plate cone test results. The equivalent SPF strains for these specimens are significantly lower than the butt welded counterparts. The most significant difference between the two sets of the results is the thickness of the material. The bead on plate material is only 1.2mm thick. The tool used to create the 1.6mm thick welds has been modified for use on this material.

The probe length has been altered but the rest of the geometry remains the same effectively increasing the ratio of probe length to probe diameter. The use of this tool on this material has not been very effective, producing welds that fall well short of the 200% equivalent strain required to be deemed superplastic.

114

Table 7.9, AA2004 Bead on Plate Weld Cone Test Results.

It should be noted that the pole thickness is the same as those produced in the 1.6mm material suggesting that this is the limit for the FSW material created using this tool.

During the SPF process the microstructure is undergoing intense deformation and at the same time microstructural change including grain growth. When grains grow the strength of the material is reduced until failure occurs. Regardless of the original thickness of the material the pole thicknesses for the AA2004 FSWs fall in the range of ~0.3 to 0.5mm.

7.5 Summary

This chapter has described the results from mechanical property tests and formability tests carried out on AA5083 and AA2004 friction stir welds in butt weld and bead on plate configurations. Details of welding forces, spindle torque and heat input is included with visual observations to assist the assessment of the welds themselves and the quality of the resulting microstructures for subsequent SPF operations.

The results have proved that superplastic deformation is observed in friction stir welded material across a complete transverse section of the weld. That is to say, the tests include material from the weld nugget, TMAZ, HAZ and parent material. The observed results show a heterogeneous structure throughout the weld giving rise to differing levels of microstructural change within different regions of the weld.

Although superplastic deformation has been observed this is significantly limited due to the different magnitudes of structural change occurring in the weld. It is impossible to introduce a weld into the parent material without suffering from some

115

kind of heterogeneous structure which may result in premature failure of the specimen within the weld region. The weld region is an induced heterogeneity. The weld region undergoes the same microstructural changes as the parent material but at different rates and from a different starting point. The parent material undergoes a change from heavily strained, elongated grains to an equiaxed structure during the initial stages of the SPF process. For AA5083 this occurs as static recrystallization in the preheat stage; for AA2004 this occurs as dynamic recrystallization during the initial stages of forming. The FSW process produces a dynamically recrystallized weld nugget and a partially recrystallized TMAZ. This region is already in a recrystallized state and so chronologically is further into the superplastic microstructural change sequence than the parent material. For the AA5083 the recrystallized grains begin to coarsen rapidly and become undeformable. The weld material acts as an undeformable block, much like a fusion weld, which is eventually pulled apart by the deforming parent material causing a rupture type failure. The AA2004 however acts in completely the opposite way. As the weld material is already recrystallized it forgoes the transformation stage of the parent material and immediately begins to deform. The weld region forms before the parent material and is stretched to failure.